{"title":"Generative Hyper-heuristics","authors":"D. Tauritz, J. Woodward","doi":"10.1145/3583133.3595033","DOIUrl":"https://doi.org/10.1145/3583133.3595033","url":null,"abstract":"","PeriodicalId":422029,"journal":{"name":"Proceedings of the Companion Conference on Genetic and Evolutionary Computation","volume":"36 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-07-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115588370","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
For many real-world decision-making tasks, a key feature is decision explainability. Hence, the so-called glass-box models offer full explainability and are still prevalent. An important area of application is the classification of imbalanced data. We require that the proposed classifiers not make errors on the minority class while minimizing errors on the majority class. This paper proposes a method for preprocessing imbalanced data by generating minority class objects. We use a multi-criteria optimization method (NSGA-II) to avoid optimizing a single aggregate criterion. The method returns a group of non-dominated solutions from which the end user can choose the best solution from his point of view. The automatic solution selection from a Pareto front is also proposed for comparison purposes. The proposed method returns good-quality classifiers, often surpassing the quality of baseline single-objective methods, and is additionally characterized by full interpretability.
{"title":"Optimized hybrid imbalanced data sampling for decision tree training","authors":"Weronika Węgier, Michał Koziarski, Michal Wozniak","doi":"10.1145/3583133.3590702","DOIUrl":"https://doi.org/10.1145/3583133.3590702","url":null,"abstract":"For many real-world decision-making tasks, a key feature is decision explainability. Hence, the so-called glass-box models offer full explainability and are still prevalent. An important area of application is the classification of imbalanced data. We require that the proposed classifiers not make errors on the minority class while minimizing errors on the majority class. This paper proposes a method for preprocessing imbalanced data by generating minority class objects. We use a multi-criteria optimization method (NSGA-II) to avoid optimizing a single aggregate criterion. The method returns a group of non-dominated solutions from which the end user can choose the best solution from his point of view. The automatic solution selection from a Pareto front is also proposed for comparison purposes. The proposed method returns good-quality classifiers, often surpassing the quality of baseline single-objective methods, and is additionally characterized by full interpretability.","PeriodicalId":422029,"journal":{"name":"Proceedings of the Companion Conference on Genetic and Evolutionary Computation","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-07-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124404154","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Roman Kalkreuth, Z. Vašíček, Jakub Husa, Diederick Vermetten, Furong Ye, Thomas Bäck
Just over a decade ago, the first comprehensive review on the state of benchmarking in Genetic Programming (GP) analyzed the mismatch between the problems that are used to test the performance of GP systems and real-world problems. Since then, several benchmark suites in major GP problem domains have been proposed over time, which were able to fill some of the major gaps. In the framework of the first review about the state of benchmarking in GP, logic synthesis was classified as one of the major GP problem domains. However, a diverse and accessible benchmark suite for logic synthesis is still missing in the field of GP. In this work, we take a first step towards a benchmark suite for logic synthesis that covers different types of Boolean functions that are commonly used for the evaluation of GP systems. We also present baseline results that have been obtained by former work and in our evaluation experiments by using Cartesian Genetic Programming.
{"title":"Towards a General Boolean Function Benchmark Suite","authors":"Roman Kalkreuth, Z. Vašíček, Jakub Husa, Diederick Vermetten, Furong Ye, Thomas Bäck","doi":"10.1145/3583133.3590685","DOIUrl":"https://doi.org/10.1145/3583133.3590685","url":null,"abstract":"Just over a decade ago, the first comprehensive review on the state of benchmarking in Genetic Programming (GP) analyzed the mismatch between the problems that are used to test the performance of GP systems and real-world problems. Since then, several benchmark suites in major GP problem domains have been proposed over time, which were able to fill some of the major gaps. In the framework of the first review about the state of benchmarking in GP, logic synthesis was classified as one of the major GP problem domains. However, a diverse and accessible benchmark suite for logic synthesis is still missing in the field of GP. In this work, we take a first step towards a benchmark suite for logic synthesis that covers different types of Boolean functions that are commonly used for the evaluation of GP systems. We also present baseline results that have been obtained by former work and in our evaluation experiments by using Cartesian Genetic Programming.","PeriodicalId":422029,"journal":{"name":"Proceedings of the Companion Conference on Genetic and Evolutionary Computation","volume":"6 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-07-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121808591","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
This Hot-off-the-Press paper summarizes our recently published work, "Matchmaker, Matchmaker, Make Me a Match: Geometric, Variational, and Evolutionary Implications of Criteria for Tag Affinity" [8]. This work appeared in Genetic Programming and Evolvable Machines. Genetic programming systems commonly use tag matching to decide interactions between system components. However, the implications of criteria used to determine affinity between tags with respect evolutionary dynamics have not been directly studied. We investigate differences between tag-matching criteria with respect to geometric constraint and variation generated under mutation. In experiments, we find that tag-matching criteria can influence the rate of adaptive evolution and the quality of evolved solutions. Better understanding of the geometric, variational, and evolutionary properties of tag-matching criteria will facilitate more effective incorporation of tag matching into genetic programming systems. By showing that tag-matching criteria influence connectivity patterns and evolutionary dynamics, our findings also raise fundamental questions about the properties of tag-matching systems in nature.
{"title":"Tag Affinity Criteria Influence Adaptive Evolution","authors":"M. Moreno, Alexander Lalejini, C. Ofria","doi":"10.1145/3583133.3595834","DOIUrl":"https://doi.org/10.1145/3583133.3595834","url":null,"abstract":"This Hot-off-the-Press paper summarizes our recently published work, \"Matchmaker, Matchmaker, Make Me a Match: Geometric, Variational, and Evolutionary Implications of Criteria for Tag Affinity\" [8]. This work appeared in Genetic Programming and Evolvable Machines. Genetic programming systems commonly use tag matching to decide interactions between system components. However, the implications of criteria used to determine affinity between tags with respect evolutionary dynamics have not been directly studied. We investigate differences between tag-matching criteria with respect to geometric constraint and variation generated under mutation. In experiments, we find that tag-matching criteria can influence the rate of adaptive evolution and the quality of evolved solutions. Better understanding of the geometric, variational, and evolutionary properties of tag-matching criteria will facilitate more effective incorporation of tag matching into genetic programming systems. By showing that tag-matching criteria influence connectivity patterns and evolutionary dynamics, our findings also raise fundamental questions about the properties of tag-matching systems in nature.","PeriodicalId":422029,"journal":{"name":"Proceedings of the Companion Conference on Genetic and Evolutionary Computation","volume":"71 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-07-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126103522","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
This paper introduces a symbolic regression based filter transform for convolutional neural network using CGP (Cartesian Genetic Programming). Symbolic regression is a powerful technique to discover analytic equations that describe data, which can lead to explainable models and the ability to predict unseen data. In contrast, neural networks have achieved amazing levels of accuracy on image recognition and natural language processing tasks, but they are often seen as black-box models that are difficult to interpret and typically extrapolate poorly. symbolic regression approaches to deep learning are underexplored.
{"title":"Toward Symbolic Regression based Model Transform for Convolutional Neural Network","authors":"Kisung Seo, Seok-Beom Roh, Soon-Joe Gwon","doi":"10.1145/3583133.3596942","DOIUrl":"https://doi.org/10.1145/3583133.3596942","url":null,"abstract":"This paper introduces a symbolic regression based filter transform for convolutional neural network using CGP (Cartesian Genetic Programming). Symbolic regression is a powerful technique to discover analytic equations that describe data, which can lead to explainable models and the ability to predict unseen data. In contrast, neural networks have achieved amazing levels of accuracy on image recognition and natural language processing tasks, but they are often seen as black-box models that are difficult to interpret and typically extrapolate poorly. symbolic regression approaches to deep learning are underexplored.","PeriodicalId":422029,"journal":{"name":"Proceedings of the Companion Conference on Genetic and Evolutionary Computation","volume":"61 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-07-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126602101","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Michal Pluhacek, Anezka Kazikova, T. Kadavy, Adam Viktorin, R. Šenkeřík
In this paper, we investigate the potential of using Large Language Models (LLMs) such as GPT-4 to generate novel hybrid swarm intelligence optimization algorithms. We use the LLM to identify and decompose six well-performing swarm algorithms for continuous optimization: Particle Swarm Optimization (PSO), Cuckoo Search (CS), Artificial Bee Colony (ABC), Grey Wolf Optimizer (GWO), Self-Organizing Migrating Algorithm (SOMA), and Whale Optimization Algorithm (WOA). We leverage GPT-4 to propose a hybrid algorithm that combines the strengths of these techniques for two distinct use-case scenarios. Our focus is on the process itself and various challenges that emerge during the use of GPT-4 to fulfill a series of set tasks. Furthermore, we discuss the potential impact of LLM-generated algorithms in the metaheuristics domain and explore future research directions.
{"title":"Leveraging Large Language Models for the Generation of Novel Metaheuristic Optimization Algorithms","authors":"Michal Pluhacek, Anezka Kazikova, T. Kadavy, Adam Viktorin, R. Šenkeřík","doi":"10.1145/3583133.3596401","DOIUrl":"https://doi.org/10.1145/3583133.3596401","url":null,"abstract":"In this paper, we investigate the potential of using Large Language Models (LLMs) such as GPT-4 to generate novel hybrid swarm intelligence optimization algorithms. We use the LLM to identify and decompose six well-performing swarm algorithms for continuous optimization: Particle Swarm Optimization (PSO), Cuckoo Search (CS), Artificial Bee Colony (ABC), Grey Wolf Optimizer (GWO), Self-Organizing Migrating Algorithm (SOMA), and Whale Optimization Algorithm (WOA). We leverage GPT-4 to propose a hybrid algorithm that combines the strengths of these techniques for two distinct use-case scenarios. Our focus is on the process itself and various challenges that emerge during the use of GPT-4 to fulfill a series of set tasks. Furthermore, we discuss the potential impact of LLM-generated algorithms in the metaheuristics domain and explore future research directions.","PeriodicalId":422029,"journal":{"name":"Proceedings of the Companion Conference on Genetic and Evolutionary Computation","volume":"38 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-07-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122506692","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Introduction to Quantum Optimization","authors":"A. Moraglio, F. Chicano","doi":"10.1145/3583133.3595040","DOIUrl":"https://doi.org/10.1145/3583133.3595040","url":null,"abstract":"","PeriodicalId":422029,"journal":{"name":"Proceedings of the Companion Conference on Genetic and Evolutionary Computation","volume":"28 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-07-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122570667","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
This paper introduces two enhancements that apply to evolution strategies such as Augmented Random Search (ARS). These improvements target generalizable tasks with widely varying initial conditions, such as legged robot locomotion where the robot starts off in a random joint configuration. The first innovation builds upon the mirrored sampling feature of ARS. It mitigates the detrimental effect of unexplained variance on training stability by forcing the simulator to use the same random seed for both mirrored pairs. The second innovation is a multi-phase end tournament procedure performed right after the ARS method is complete. This tournament helps to ensure that the final product of training, a single model selected from the population, performs well over a wide range of random initial conditions. Improved results are demonstrated using MuJoCo simulations of legged robots.
{"title":"Evolution Strategies with Seed Mirroring and End Tournament","authors":"S. Soleyman, Joshua Fadaie, Fan Hung, D. Khosla","doi":"10.1145/3583133.3590541","DOIUrl":"https://doi.org/10.1145/3583133.3590541","url":null,"abstract":"This paper introduces two enhancements that apply to evolution strategies such as Augmented Random Search (ARS). These improvements target generalizable tasks with widely varying initial conditions, such as legged robot locomotion where the robot starts off in a random joint configuration. The first innovation builds upon the mirrored sampling feature of ARS. It mitigates the detrimental effect of unexplained variance on training stability by forcing the simulator to use the same random seed for both mirrored pairs. The second innovation is a multi-phase end tournament procedure performed right after the ARS method is complete. This tournament helps to ensure that the final product of training, a single model selected from the population, performs well over a wide range of random initial conditions. Improved results are demonstrated using MuJoCo simulations of legged robots.","PeriodicalId":422029,"journal":{"name":"Proceedings of the Companion Conference on Genetic and Evolutionary Computation","volume":"9 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-07-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"117021662","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
We address the problem of promoting diversity in online embodied evolution of heterogeneous robot swarms. We argue that it is not easy to adapt existing diversity algorithms from traditional evolutionary robotics to this context and describe a method in which selection is based on originality and which allows a swarm of heterogeneous agents to maintain a high degree of diversity in behavioral space. We also describe a behavioral distance measure that compares behaviors in the same conditions to provide reliable measurements in online and distributed contexts. We test the selection scheme on an open-ended survival task and show its effectiveness. Without any other pressure besides that of the environment, the evolved strategies tend toward simplicity, exploiting the existing affordances. An additional external pressure enables the emergence of rich and diverse behaviors.
{"title":"Promoting Originality in Online Swarm Robotics","authors":"Amine M. Boumaza","doi":"10.1145/3583133.3590563","DOIUrl":"https://doi.org/10.1145/3583133.3590563","url":null,"abstract":"We address the problem of promoting diversity in online embodied evolution of heterogeneous robot swarms. We argue that it is not easy to adapt existing diversity algorithms from traditional evolutionary robotics to this context and describe a method in which selection is based on originality and which allows a swarm of heterogeneous agents to maintain a high degree of diversity in behavioral space. We also describe a behavioral distance measure that compares behaviors in the same conditions to provide reliable measurements in online and distributed contexts. We test the selection scheme on an open-ended survival task and show its effectiveness. Without any other pressure besides that of the environment, the evolved strategies tend toward simplicity, exploiting the existing affordances. An additional external pressure enables the emergence of rich and diverse behaviors.","PeriodicalId":422029,"journal":{"name":"Proceedings of the Companion Conference on Genetic and Evolutionary Computation","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-07-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129599517","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
In complex network systems, the problem that how to select members with considerable information-spreading ability, i.e., the influence maximization (IM) problem, is a current research hotspot. In practice, networked systems are extremely vulnerable to interferences from external sources or even human sabotages, which cause direct disturbances on the topology. One of the common attacks is cascading failures. To cope with the IM problem under cascading failures, a new metric RS-cf is defined to evaluate the performance of seeds under this attack model. Guided by this, a Memetic algorithm, named MA-RIMcf, is devised to determine those nodes with both robustness and influential ability. The reasonableness and effectiveness of the algorithm are verified by experiments on synthetic network data. These solutions are expected to solve the influence maximization problem in realistic environments.
{"title":"A Memetic Algorithm to solve the Robust Influence Maximization Problems against Cascading Failures","authors":"Shun Cai, Shuai Wang, Zhaoxi Ou","doi":"10.1145/3583133.3590615","DOIUrl":"https://doi.org/10.1145/3583133.3590615","url":null,"abstract":"In complex network systems, the problem that how to select members with considerable information-spreading ability, i.e., the influence maximization (IM) problem, is a current research hotspot. In practice, networked systems are extremely vulnerable to interferences from external sources or even human sabotages, which cause direct disturbances on the topology. One of the common attacks is cascading failures. To cope with the IM problem under cascading failures, a new metric RS-cf is defined to evaluate the performance of seeds under this attack model. Guided by this, a Memetic algorithm, named MA-RIMcf, is devised to determine those nodes with both robustness and influential ability. The reasonableness and effectiveness of the algorithm are verified by experiments on synthetic network data. These solutions are expected to solve the influence maximization problem in realistic environments.","PeriodicalId":422029,"journal":{"name":"Proceedings of the Companion Conference on Genetic and Evolutionary Computation","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-07-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129814038","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：